(dcw3, dcw4, dcw5) - university of auckland d… · (dcw3, dcw4, dcw5) •no. •research domain...
Post on 19-Jul-2020
0 Views
Preview:
TRANSCRIPT
Growing Up in New Zealand
4 Year External Data Release
(DCW3, DCW4, DCW5)
12 September 2017
Susan Morton, Sarah Berry, Caroline WalkerAvinesh Pillai, Peter Tricker
University of Auckland
www.growingup.co.nz
Outline
1. Study overview
2. Focus of current release – four year data
3. Growing Up In New Zealand external data
4. Applying for external data
5. Questions
Overarching Aim of Growing Up in New Zealand
To provide contemporary population relevant evidence about
the determinants of developmental trajectories for 21st
century New Zealand children in the context of their families.
“The Ministry of Social Development and the Health Research
Council of New Zealand, in association with the Families
Commission, the Ministries of Health and Education and the
Treasury, wish to establish a new longitudinal study of New
Zealand children and families, ….” to gain a better
understanding of the causal pathways that lead to particular
child outcomes (across the life course)
…… introduction to RFP in 2004.
New Zealand’s contemporary longitudinal study
Conceptual framework for child development Growing Up in New Zealand
• Life course approach
• Child centred
• Multi-disciplinary
• Dynamic interactions
• Change over time
• Understanding trajectories
• Intergenerational
• Understanding environmental
influences (proximal and distal)
• Biology and social contexts
• Putting the “environment into the
epigenetic”
Shulruf, Morton et al. (2007) Eval & Hlth Prof 30:2017-28
The Growing Up in New Zealand cohort
• Recruited 6853 children before their birth - via
pregnant mothers (6823)
• Partners recruited and interviewed
independently in pregnancy (4401)
• Cohort has adequate explanatory power to
consider trajectories for Maori (1in 4), Pacific (1
in 5) and Asian (1 in 6) children, and to consider
multiple ethnic identities (approx. 40%)
• Cohort broadly generalisable to current NZ births
(diversity of ethnicity and family SES)
• Retention rates to 4 year DCW have been very
high (over 90% of antenatal)
Longitudinal Information during pre-school period
Child age Ante-
natal
Peri-
natal
6
W
35
W
9
M
12
M
16
M
23
M
2
Y
31
M
45
M
54
M
Mother
CAPI*✓ ✓ ✓ ✓
Father CAPI*✓ ✓ ✓
Mother
CATI†✓ ✓ ✓ ✓ ✓
✓
Child‡✓ ✓ ✓
Data
linkage**✓ ✓ ✓ ✓
* CAPI computer assisted personal interview
† CATI computer assisted telephone interview
‡ Child measurements
** Linkage to routine health records
Each DCW represents a snapshot of development
Partnerships to facilitate translation
Study design
Data collection
Data analyses
Dissemination of results
Policy interaction
Reporting: following each data collection wave study reports present key findings
Policy interactionPolicy forum: representatives from 16 key government agencies. Advice on specific priorities for data collection, data analysis. Develop collaborative evaluation projects.
Data linkage: Opportunities for l inkage to routine Health, Education and Social BiG Datasets (with informed consent)
Policy interactionPolicy forum: advice on policy priorities for data analyses
and for timely and relevant reporting
Policy interactionPolicy briefs: opportunities to provide evidence to policy submission processes. Minister/Ministerial questions answered
Data Access: Opportunities for fast-track, bespoke reports, external data access to datasets
Moving beyond “risk factorology”
Hearing from the children and the families directly to understand WHY we
see associations, WHAT WORKS, WHEN, and for WHOM.
External Data Release – Preschool data collections
Retention to 4
Parental antenatalinterview
6 weeks
9 month interview
2 year interview
45 month ca ll
54 month interview
Pregnant mothers N = 6822 *
Partners N = 4401
Chi ld counts (N = 6853)
Completed = 6843 Skipped = 10
Chi ld counts (N = 6795)
Completed = 6476 (94%) Skipped = 310
Lost to follow up = 9
Opt out =54Deceased =4
Chi ld counts (N = 6706) Completed = 6327 (92%)
Skipped = 366Lost to follow up = 13
Chi ld counts (N = 6670)
Completed = 6207 (91%) Skipped = 442
Lost to follow up = 21
Chi ld counts (N = 6639)
Completed = 6156 (90%)
Skipped = 462Lost to follow up = 21
Opt out = 88
Deceased = 1
Opt out = 36
Opt out = 29Deceased = 2
4 year data collection - key measures
Data Life Cycle
Centralised repositoryGrowing Up in New Zealand data is centrally
collated, cleaned, audited and managed.
Analytical data preparationGrowing Up in New Zealand data for
a data collection wave is prepared for analysis
Data anonymisationGrowing Up in New Zealand data is
prepared for external release
Raw data
Code text data
Cleaning
Derived information
Merge data
Formats & label assignment
Order variables
Create final data set
Analytical data preparationGrowing Up in New Zealand data for
a data collection wave is prepared for analysis
Sources of data
– Mother (M): information about the GUiNZ child’s
mother and her household
– Partner (P): information about partner of
GUiNZ child’s mother & their household
– Child Proxy Mother (CM):information about the GUiNZ
child provided by their mother
– Child Proxy Partner (CP):information about the GUiNZ
child provided by mother’s partner
– Child Observation (CO): information about the GUiNZ
child collected by the interviewer
Linkage, scales, tools and observations
• Linkage data
– Immunisation register
– Respiratory hospitalisation and admission
• Scales or tools
– Strengths and difficulties questionnaire
• Child observations
– Anthropometry
- Weight, height, waist circumference
Longitudinal datasets
Naming convention
Data collection wave Full dataset nameShort name for the
dataset
Variable suffixReference for variable suffix
DCW0Antenatal Mother DCW0M _AM Antenatal Mother
Antenatal Partner DCW0P _AP Antenatal Partner
DCW1
Nine month child dataset
DCW1C
_W6 Six week call
_PDL Perinatal
_M9CM Nine month child
Nine month mother dataset
DCW1M _M9M Nine month mother
Nine month partner dataset
DCW1P _M9P Nine month partner
DCW2
Two year child dataset DCW2C
_M16CM Sixteen month child
_M23CM Twenty three month child
_Y2CM Two year child
Two year mother dataset
DCW2M
_M16M Sixteen month mother
_M23MTwenty three month mother
_Y2M Two year mother
Two year partner dataset
DCW2P _Y2P Two year partner
DCW331M child & mother dataset
DCW3C_M31CM 31 month child
_M31M 31 month mother
DCW445M child dataset DCW4C _M45CM 45 month child
45M mother dataset DCW4M _M45M 45 month mother
DCW554M child dataset DCW5C _M54CM 54 mother child
54M mother dataset DCW5M _M54M 54 month mother
Focus of current release
Current data release – using the data
• Longitudinal data
– 13 datasets in total…which time point(s) is
the focus?
– Combining longitudinal items (merging
datasets. Do I have the correct
denominator?
• Longitudinal data
– Missing data across time points
– Different denominators if merging datasets
across time
– Be aware of answers such as ‘98’ (refused to
answer) and ’99’ (don’t know), when
calculating summaries.
Current data release – using the data
Current data release – using the data
- Individual data not aggregated data
- Data storage security considerations
- Software considerations
- Access to datasets
- Identification keys provide the relationships
between the datasets
- Child to Child relationships
- Child to Mother/ Partner relationships
- Mother to Partner relationships
Data documentation
Reference and Process User Guide
Questionnaires
Data Dictionaries
❖ Growing Up in New Zealand Questionnaires and Data dictionaries are/ will be available online
Data dictionary fields (DCW3, DCW4, DCW5)
• No.
• Research Domain
• Subdomain
• Questionnaire number
• Question
• Variable name in external dataset
• Formatted data values
• Variable Type
• Notes
❖ Growing Up in New Zealand Questionnaires and Data dictionaries are available online
1. Identification key
2. Raw Variables
3. Categorised Variables
4. Re-classified Variables
5. Derived Variable
Data dictionary fields (DCW3, DCW4, DCW5)
Using the Growing Up in New Zealand DataA researchers perspective
• Research Question
– Example: Is parental gambling in the first 9 months of life associated with health outcomes at 2 and 4 years
• Data sources
– Mother and Partner (Antenatal and 9 month)
- Gambling variables
- Covariates: Ethnicity, education, age, deprivation, income
– Child proxy (2 and 4 years)
- Outcome: Parent report health
- Outcome: Strengths and difficulties scale
– Linkage
- Outcome: Hospital admission data
- Outcome: Immunisation register
Variables – sources and coding
• Merge dataset
– Family: Mother and Partner gambling
– Child: Outcomes
• Single response or multiple response
– Example: Gambling variable
• Coding
– Binary coding: 0 or 1
– Numeric coding: 1, 2, 3, 4
Variables – sources and coding
• Derived variable
– Create a derived variable from multiple response options
– Ethnicity
- Externally prioritised or self prioritised
• Scales
– Strengths and difficulties questionnaire
• Linkage
– Respiratory illness admitted to hospital
Further information
www.growingup.co.nz
Further information
dataaccess@growingup.co.nz
Data dictionaries
www.growingup.co.nz
Accessing the data
Externalresearcher
Attend a workshop
Completean application
Accessgranted
DAC process
Remote Data Access Platform
Data Access Protocol
• Principles governing the data access process
• Application process
• Safeguarding the privacy of participants
• Ensuring the sustainability of the project
• Role and function of the Data Access Committee
• Authorship and publications
Other ways to access the data
Growing Up in
New Zealand team
Accredited
researcher
Bespoke report
Acknowledgements
• Children and their families
• UoA Growing Up team
• Superu (lead agency)
• Ministry of Social Development
• Multiple government agencies
• Policy Forum
• Advisory and Stakeholder groups
• All funders (government)
Questions
top related